Wireless network resource allocation method and device for integrated energy service station
By establishing a network system model in mobile edge computing and formulating different resource allocation strategies, the problems of high cost of resource allocation and management and poor economic scalability in mobile edge computing are solved, and the efficiency of edge nodes and wireless access network resources are achieved. , fair distribution.
Patent Information
- Application Number
- CN202110715592.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-06-25
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2041-06-25
AI Technical Summary
There are problems of high cost and poor economic scalability in resource allocation and management in mobile edge computing, especially due to the limited capacity of edge nodes and wireless access networks, the heterogeneity of resource and service requirements is difficult to effectively solve.
A comprehensive energy service station wireless network resource allocation method is proposed. By establishing a network system model, formulating market balance strategy, social optimal strategy or proportional sharing strategy, optimizing resource allocation of edge nodes and wireless units, ensuring efficient resources under budget constraints , distribution fairly.
It realizes efficient and fair distribution of edge nodes and wireless access network resources, reduces resource allocation costs, maintains the economic scalability of mobile edge computing, and solves the heterogeneity of resource and service requirements.
Smart Images

Figure CN113923781B_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the technical field of data resource allocation, and in particular to a wireless network resource allocation and device for an integrated energy service station. Background Art
[0002] The advent of cloud computing (CC) has dictated the centralization of computing power into large data centers. As this trend is expected to continue in the future, cloud computing will play an increasingly prominent role in web-based services by providing supercomputing-like capabilities. However, the distance between data centers and end users is often long, so cloud computing is not suitable for new time-sensitive use cases, such as the fifth generation of mobile radio networks (5G).
[0003] Mobile Edge Computing (MEC), introduced in existing technologies, promises to surpass this inherent limitation of cloud computing by shifting computing power to the edge of the network (close to the Radio Access Network (RAN)). The resulting reduction in propagation delay promotes a wide range of time-sensitive applications such as the Tactile Internet, Augmented Reality, and the Internet of Things. In addition, MEC can reduce the burden that future data-intensive applications (such as autonomous driving and massive IoT) may bear on the network infrastructure, as local computing does not require sending large amounts of data to the cloud.
[0004] MEC is currently at a development stage and among the different challenges it poses to the research community, resource allocation and management is an important one. Unlike the usual case in cloud computing resource management research, unlimited network capacity between computing nodes and offloading sources is less than ideal. In fact, edge nodes (ENs) are deployed in the vicinity of wireless access points with limited capacity. Moreover, given the relatively high density of these deployments, the computing power provided by edge nodes is limited.
[0005] Therefore, how to reduce the cost of market-oriented network resource allocation while maintaining the economic scalability of mobile edge computing has become a technical problem that needs to be solved urgently. Summary of the invention
[0006] In view of this, the purpose of the present invention is to propose a method and device for allocating wireless network resources of an integrated energy service station, so as to solve the technical problem of how to reduce the cost of allocating wireless network resources of an integrated energy service station while maintaining the economic scalability of mobile edge computing.
[0007] Based on the above objectives, a first aspect of the present disclosure provides a method for allocating wireless network resources of an integrated energy service station, comprising:
[0008] At least one user terminal transmits a data processing task to a network system model pre-constructed by connecting multiple edge nodes and multiple wireless units, wherein the edge nodes are network resources corresponding to data calculation and processing, and the wireless units are wirelessly connected to at least one server terminal and at least one user terminal respectively;
[0009] The at least one server allocates appropriate edge nodes for data processing for each data processing task in the network system model according to the resource allocation strategy, wherein the predetermined resource classification strategy includes: market equilibrium strategy, social optimal strategy or proportional sharing strategy.
[0010] A second aspect of the present disclosure provides a wireless network resource allocation device for an integrated energy service station, comprising:
[0011] A data transmission module, used for at least one user end to transmit a data processing task to the network system model pre-constructed by connecting multiple edge nodes and multiple wireless units, wherein the edge nodes are network resources corresponding to data calculation and processing, and the wireless units are wirelessly connected to at least one server end and at least one user end respectively;
[0012] The data processing module is used for allocating appropriate edge nodes for data processing to each data processing task in the network system model according to the resource allocation strategy by the at least one server, wherein the predetermined resource classification strategy includes: market equilibrium strategy, social optimal strategy or proportional sharing strategy.
[0013] A third aspect of the present disclosure provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the first aspect when executing the program.
[0014] A fourth aspect of the present disclosure provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute any method described in the first aspect.
[0015] From the above, it can be seen that the integrated energy service station wireless network resource allocation method and device provided by the present disclosure can formulate different resource allocation strategies in advance after the network system model is established, specifically market equilibrium strategy, social optimal strategy or proportional sharing strategy, so that the service end can allocate appropriate edge nodes for data processing for each data processing task in the network system model according to the market equilibrium strategy, social optimal strategy or proportional sharing strategy. This integrated energy service station wireless network resource allocation method allows the service end to allocate network resources according to different resource allocation strategies, ensuring that the wireless unit and computing resources (i.e., edge nodes) can perform their related data processing tasks while being subject to budget constraints, thereby solving the problem of limited capacity of edge nodes and wireless access networks and the heterogeneity of resources and service requirements, and can maximize fairness, while ensuring that all service ends obtain the network resources they need, thereby achieving efficient and fair network resource allocation management. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the present disclosure or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0017] Figure 1 A flowchart of a method for allocating wireless network resources in an integrated energy service station according to an embodiment of the present disclosure;
[0018] Figure 2 A schematic diagram of a network system model constructed in an embodiment of the present disclosure;
[0019] Figure 3 It is a structural block diagram of a wireless network resource allocation device for an integrated energy service station according to an embodiment of the present disclosure;
[0020] Figure 4 It is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0021] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.
[0022] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present disclosure should be understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprising" and similar words mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects.
[0023] In related technologies, the emerging concept of network slicing is well suited to provide the required heterogeneous virtualized resource bundles. However, the slice functionality needs to be customized according to the services desired by the slice owner. Different slices may require different amounts of computing and communication resources. In addition, the underlying shared infrastructure is highly heterogeneous. Not only are there huge differences in the availability and utility of resources between the MEC and RAN domains, but edge nodes may present different installed capacities in a single MEC cluster.
[0024] This huge diversity requires co-optimization of resources in both domains, as decoupling the two domains can lead to a situation where resources in one domain are over-provisioned due to capacity in the other domain. For example, if one slice receives a large amount of radio capacity without sufficient computing resources, most jobs uploaded over the RAN will experience unacceptable processing delays. At the same time, other slices with less stringent processing requirements can more efficiently utilize the over-provisioned radio resources.
[0025] Indeed, the inherent spatial and temporal heterogeneity in demand and resource availability suggests that flexible allocations may yield greater efficiency and better performance.
[0026] Based on the above mentioned situation, the following problem arises: Considering the MEC / RAN deployment consisting of a set of heterogeneous network elements and edge nodes, how to effectively allocate computing and network resources (i.e., edge nodes) to competing network slices with different requirements while ensuring fairness, service priority, and economic convenience becomes a problem that needs to be solved.
[0027] In the related art, a balancing method is used to allocate the edge node resources with heterogeneous capacity constraints to multiple competing services at the edge of the network by reasonably pricing edge nodes with different geographical distributions, thereby improving the computing resource utilization of the edge nodes. However, this scheme only studies the system from a high-level perspective of the edge computing platform (i.e., without considering the radio access part). This assumption is not ideal because future services, whether through large workloads (i.e., video transcoding, image processing), a large number of simultaneous requests (i.e., large-scale Internet of Things), or a combination of the two (i.e., autonomous driving and connected cars), will challenge the radio access part of the network.
[0028] In the related technology, an iterative algorithm is used to converge to the local optimal solution of the original non-convex problem based on a new successive convex approximation technique, thereby achieving optimal management of radio resources and computing resources. The optimization aims to minimize the energy consumption of users while meeting the delay constraints. Although this method manages radio resources and computing resources in a heterogeneous network in an integrated manner, it cannot guarantee fairness, service priority, and economic convenience in the resource management allocation process because it calculates the global optimal solution of both through an algorithm.
[0029] Among the related technologies, some use market auction models to configure different service qualities for different network slices. Although this solution better solves the management of wireless resources from a market perspective, it only considers resource allocation in wireless communication networks, and does not involve resource management in edge computing networks.
[0030] Based on the above situation, the wireless network resource allocation method of the integrated energy service station of the present disclosure is as follows: Figure 1 As shown, including:
[0031] Step 100, at least one user terminal transmits a data processing task to a network system model pre-constructed by connecting multiple edge nodes and multiple wireless units, wherein the edge nodes are network resources corresponding to data calculation and processing, and the wireless units are wirelessly connected to at least one server terminal and at least one user terminal respectively.
[0032] In this step, the server side (i.e., at least one server of the service provider that performs data processing) obtains the corresponding edge node as a network resource through the network system model, uses the obtained network resources to communicate through the wireless unit in the network system model, processes the data processing task of the user side, and completes the data processing task.
[0033] Step 200, at least one server allocates appropriate edge nodes for data processing for each data processing task in the network system model according to a resource allocation strategy, wherein the predetermined resource classification strategy includes: a market equilibrium strategy, a social optimal strategy or a proportional sharing strategy.
[0034] Among them, the market equilibrium strategy, that is, taking the edge nodes of the entire network system model as resources and setting corresponding values for each edge node, so that the server, as a buyer, allocates resources in accordance with the market buying and selling method so that the entire network system model is in a state of balanced resource allocation. This allocation method can ensure that each data processing task can be processed evenly and effectively after allocation.
[0035] Among them, the socially optimal strategy, that is, the sum of utilities that the server can utilize to process data using edge nodes in the network system model, can be maximized, so that network resources can be fully utilized.
[0036] Among them, the proportional sharing strategy is that the server allocates market-oriented network resources according to the corresponding resource ratio. In this way, more network resources can be allocated to servers with larger data processing capabilities, ensuring that each server can be allocated network resources that match its actual resource ratio.
[0037] Through the above scheme, different resource allocation strategies can be formulated in advance after the network system model is established, specifically market equilibrium strategy, social optimal strategy or proportional sharing strategy, so that the server can allocate appropriate edge nodes for data processing for each data processing task in the network system model according to the market equilibrium strategy, social optimal strategy or proportional sharing strategy. This integrated energy service station wireless network resource allocation method allows the server to allocate network resources according to different resource allocation strategies, ensuring that wireless units and computing resources (i.e., edge nodes) can perform their related data processing tasks while being subject to budget constraints, thereby solving the problem of limited capacity of edge nodes and wireless access networks and the heterogeneity of resources and service requirements, and can maximize fairness while ensuring that all servers obtain the network resources they need, thereby achieving efficient and fair network resource allocation management.
[0038] In a specific embodiment, the resource allocation strategy is a market equilibrium strategy, and step 200 specifically includes:
[0039] Step 211, determine the utility data u of each server s , and the resource budget value B required for the server to process each data processing task s .
[0040] In this step, the utility data of a server is the utility situation of the server. The resource budget value of a server is the resource budget value required by the server to process the corresponding data processing task, that is, the amount of resources that the server needs to purchase as a buyer to process the corresponding data processing task.
[0041] Step 212: determine for each data processing task the first objective function. The corresponding edge nodes allocate data so that the corresponding data processing tasks can be processed using the edge node allocated data, where s represents each server and S is the total number of servers.
[0042] In this step, two conditions need to be met: first, to ensure that each service provider can obtain the most valuable resources; second, the edge nodes in the network system model have been fully allocated (i.e., all resources are fully utilized) or the value of the edge nodes corresponding to the data processing task is zero (i.e., the task is fully allocated and processed). The basis for judging whether these two conditions can be met is to judge whether the market-based network resource allocation situation corresponding to each service end satisfies the above-mentioned first objective function. If it does, it proves that the above two conditions are met, and the corresponding edge node allocation data in the corresponding market-based network resource allocation situation is used to perform data processing on the corresponding data processing task.
[0043] In this way, the network system model can always be in a state of balanced distribution.
[0044] In a specific embodiment, step 211 specifically includes:
[0045] Step 2111: Determine the amount of work that each server is executing simultaneously on the edge node side
[0046] In this step, we first obtain the computing resource capacity r required by the server to complete the corresponding data processing task and the corresponding wireless unit c for the server's payload upload resource capacity
[0047] Then, obtain the computing resource r capacity x of the corresponding server in the edge node m allocated to the edge node side s,m,r , and the network resource capacity y allocated to the server in the wireless unit c on the wireless unit side s,c .
[0048] Based on the above acquired data, the amount of work that the server executes simultaneously on the edge node side is determined using the formula:
[0049] Among them, X s =(x s,m,r)∈R M×R It is the computing resource allocation matrix, M is the total number of edge nodes, and R is the total amount of computing resources corresponding to one edge node.
[0050] Step 2112, determine the effective load of each server on the wireless unit side
[0051] In the above steps, based on the obtained and s,c , determine the server's payload on the radio unit side:
[0052] Where C is the total number of wireless units.
[0053] Step 2113, determining the utility data of each server based on the minimum value of the maximum workload and the payload
[0054] Step 2114: retrieve at least one resource budget score B used by each data processing task from the memory of the corresponding server. i , set at least one resource budget score B i Accumulate and obtain the resource budget value B corresponding to each data processing task s .
[0055] Through the above steps, some corresponding basic data can be obtained based on the constructed network system model, which facilitates the subsequent optimization according to the corresponding first objective function in step 212 to obtain a balanced distribution. and best server utility In this way, resources can be allocated based on the network system model.
[0056] In a specific embodiment, the resource allocation strategy is a social optimal strategy, and step 200 specifically includes:
[0057] Step 221, determine the utility data u that can be executed simultaneously on each server s .
[0058] In this step, u s The process of obtaining is the same as step 2111 to step 2113 above, and will not be repeated here.
[0059] Step 222, using the second objective function Calculate the sum of utility data of at least one server, and determine the corresponding edge node allocation data for each data processing task based on the sum of utility data, so as to use the edge node allocation data to perform data processing on the corresponding data processing task, where s represents each server and S is the total number of servers.
[0060] In the above steps, regardless of the budget differences between the service ends, the performance of the entire network system model (analogously referred to as society) can be maximized through the second objective function (social optimal function). In other words, the number of service users in the network system model can be maximized, thereby making full use of the resources of the network system model.
[0061] In a specific embodiment, it also includes:
[0062] Step 223, calculate the resource budget value B required by the server to process each data processing task s .
[0063] Step 224, after determining the resource budget value B s When it is greater than a predetermined threshold, the third objective function is used The utility data sum is calculated, and the corresponding edge node allocation data is determined for each data processing task according to the utility data sum, so that the corresponding data processing task is processed by using the edge node allocation data.
[0064] In the above steps, for some resources with large costs (i.e., resource budget value B s greater than a predetermined threshold), assign corresponding weights to these servers (i.e., the corresponding B s ), and use the third objective function to perform optimal allocation, ensuring the advantage of the corresponding service end (i.e., service provider) in the network system model, thereby making the entire network allocation more fair.
[0065] In a specific embodiment, the resource allocation strategy is a proportional sharing strategy, and step 200 specifically includes:
[0066] Step 231: Determine the workload of the target server s' on the edge node side according to the resource ratio of the target server s' And, determine the payload size of the target server s' on the wireless unit side
[0067] Step 232, determine the utility data of the target server s'
[0068] Step 233, based on the obtained utility data u of the target server s' s’ (X s’ , Y s’ ) distributes data to the target server s', so that the target server s' processes the data according to the data distribution result.
[0069] The resource allocation scheme obtained through the above steps does not consider the needs of the supplier represented by the server, nor the load of the edge nodes and wireless units. It only allocates all available resources in proportion, so that the server with a larger budget can obtain a larger share of resources. Therefore, this resource allocation scheme gives priority to those servers with larger budgets.
[0070] In a specific embodiment, step 231 specifically includes:
[0071] Step 2311, obtain the capacity of resource r in edge node m and the network resource capacity in wireless unit c
[0072] Step 2312, determine the resource budget value B of each server s .
[0073] Step 2313, calculate the resource budget value B of the target server s' s’ The proportion of:
[0074] Step 2314, determine the computing resource allocation amount of the target server s' and the market-based network resource allocation of the target server s' in, S is the total number of servers, M is the total number of edge nodes, and C is the total number of wireless units.
[0075] Step 2315, determine the workload of the target server s' on the edge node side:
[0076] Step 2316, determine the payload size of the target server s' on the wireless unit side:
[0077] Through the above steps, various basic data required for step 232 can be obtained, so that step 232 can be successfully completed.
[0078] Based on the above scheme, simulation is carried out according to the market equilibrium strategy, social optimal strategy and proportional sharing strategy based on the network system model.
[0079] The performance corresponding to each strategy is obtained based on the cumulative distribution function of the sum of concurrently executable jobs on each server. That is, the more jobs that can be executed simultaneously, the higher the performance, and the better the performance corresponding to this strategy.
[0080] In addition, the efficiency data corresponding to this strategy is calculated based on the ratio of the total number of jobs executed simultaneously by each server in the corresponding strategy to the maximum number of jobs allowed to be executed simultaneously by the system capacity. However, from the efficiency point of view, the efficiency of the social optimal strategy is higher, but the social optimal strategy is allocated according to the optimal allocation method. This allocation method may allocate very few resources to some small servers, so the fairness of this allocation method will be worse.
[0081] The specific resource allocation strategy to be selected needs to be selected based on the network system model and the actual situation of each server. These strategies can be combined and arranged arbitrarily. The specific combination and arrangement can be selected according to actual needs and is not specifically limited here.
[0082] Based on the solution described in the above embodiment, the present disclosure proposes another embodiment of a wireless network resource allocation method for an integrated energy service station, specifically:
[0083] It is divided into three parts. The first part will introduce the construction of the system model; the second part will introduce the allocation method of network resources; and the third part proposes a market-oriented network resource allocation simulation based on the constructed system model and resource allocation method.
[0084] Table 1
[0085]
[0086]
[0087] Table 1 above shows the meaning of each representative symbol, including edge node (EN), mobile edge computing (MEC), radio access network (RAN), and service provider (SP).
[0088] 1. Mobile edge computing and radio network system model (i.e., network system model)
[0089] Consider a MEC scenario where multiple ENs form a MEC cluster. Let M be the set of such nodes in the MEC cluster and be the set resource types provided by edge nodes (i.e., CPU, RAM, and storage). The available capacity of resource type r∈R is expressed in edge nodes m∈M. Since we consider a heterogeneous MEC cluster, the capacity In principle it can be different for each resource and edge node to simulate real-life scenarios where ENs have different installed capacities.
[0090] Consider a radio access network that users of the MEC system can leverage to offload work to ENs. Let C be the set of C cells (i.e., the set of radio cells) that can access the MEC cluster. We then define the available capacity of each cell c (i.e., radio cell c) ∈ C as For simplicity, we consider resources to be expressed in terms of spectrum portions of cells. We assume the existence of a slice-aware MAC scheduler that is able to dynamically enforce these restrictions, and although RAN access capacity is limited, we do not impose any restrictions on the capacity of the backhaul network connecting each cell to any server in the MEC cluster. Furthermore, once a job payload enters the cluster, it can be freely scheduled for execution at any edge node in the cluster.
[0091] Construct a network system model, such as Figure 2 shown.
[0092] To coordinate the resource allocation based on the market approach, the network system needs to consider the following two issues: demand-based service characteristics and service utility functions. The solutions to the two issues are explained below.
[0093] 1.1. Service characteristics based on demand
[0094] A service provider (SP) (i.e., a server) owns a virtual bundle of computing and radio resources in the form of a network slice. Network slices and service providers have interchangeable definitions in this article.
[0095] Service providers use these resources to provide specific MEC services to end users. For example, a specific service provider may provide a neural network training offload service. In this case, the end user needs to upload the training data to the edge node, which in turn trains the neural network, which we call a combination of upload and processing work. Alternatively, some services such as speech recognition or video processing may require end users to send a steady stream of data to the EN for processing. However, such streams usually consist of individual packets, and uploading and processing each packet may be considered work.
[0096] Independent of its scope, a particular service can usually be described in terms of the computing and network resources required to complete the associated job. End-user jobs are likely to present similar resource requirements. Therefore, we define a set of requirements, with each It is worth noting that given the wide variety of MEC applications, services can present different demand profiles: there may be CPU-intensive services whose CPU requirements may be relatively higher than their memory requirements, or network-intensive services whose work payload is larger than other payloads, thus requiring higher bandwidth allocations.
[0097] Let S be the service provider. We will is defined as the minimum number of r-type MEC resources required to complete a job of service provider s in a timely manner. Similarly, indicates that the network resources required to successfully upload the work of service provider s through unit c are minimal. It does not depend on the specific EN in which the job is performed, but must depend on the cell where the payload is uploaded. This is because different channel qualities may require higher or lower bandwidth requirements for the same payload size. In particular, by appropriately adjusting Several factors that can positively or negatively affect channel quality can be addressed. For example, can be increased in certain ranges to counteract reduced spectrum utilization due to local inter-area interference or harsh end-user propagation conditions. Furthermore, if there are large differences in channel quality among the end-users of a given service provider within a single cell c, it is always possible to partition the end-users subscribing to a single service into groups of end-users nG with quasi-homogeneous channel quality. Each group subscribes to a fictitious sub-service si (with i = 1, ..., nG) that differs from the original sub-service only in its network resource requirements.
[0098] Finally, all All consistent factors are considered as time averages so that the resource allocation mechanism can operate on a stable and reliable performance of the RAN. If network conditions change or new market allocations occur, the resource allocation mechanism can be updated. and recalculate the final distribution.
[0099] 1.2. Service Utility Function
[0100] In the multi-resource allocation scenario, one aspect that cannot be ignored is the existence of resource bottlenecks and the emergence of dominant resources. In the above scenario, when the relevant demand profile of a particular service provider exhausts the resources available in the edge node, it may encounter resource bottlenecks in a specific edge node of the cluster. It is called resource dominance. When the edge node capacity of such resources is exhausted, no more jobs can be scheduled even if other resource types are available in the EN.
[0101] On the other hand, there is no risk of encountering a RAN domain bottleneck, since the allocatable resources are of only one type. However, given the time sensitivity of the jobs, queuing within the domain is not allowed, which means that the domain with the least execution limits the total system throughput. This can also be considered a bottleneck, since once the capacity of one domain reaches the saturation point, no additional work can be accommodated in the network system. In the constructed network system model, the existence of these resource bottlenecks in the system is mathematically captured by properly defining the service provider's utility function.
[0102] In the related art, based on actual virtualization schemes, network and cloud operators rent capacity from service providers based on job processing resources and have no idea how they will use this capacity to serve end users. Instead, the evaluation of system performance must rely entirely on parameters provided by network and cloud operators. Therefore, a service utility function is defined that is transparent to the queuing strategy and admission control techniques adopted by the service provider. In order to do this, it is noted that to meet the latency requirements, at any point in time, the system can only accommodate a limited number of jobs simultaneously, regardless of the above choices. This number is considered to be a valid overall performance indicator. It explicitly determines for the SP how many end users the system can serve simultaneously for a given resource bundle.
[0103] Let x s,m,r is the capacity of computing resource type r reserved for service provider s in node m. Then, for any resource r, denotes the maximum number of concurrent jobs s that this allocation allows to execute in EN m. However, the actual number of concurrently executed jobs is limited by the primary resources. Therefore, the minimum of these numbers across all computing resource types must be considered. The maximum number of jobs s that can be executed simultaneously in the service provider's MEC domain with an acceptable per-job performance is obtained by summing over the edge nodes in the system:
[0104]
[0105] Where X s =(x s,m,r )∈R M×R is the computational resource allocation matrix.
[0106] Similarly, by adding y s,c Assume that the network resources allocated to service provider s in region c, y s,c and The ratio between them is determined as the maximum number of workloads that can be sent to service s simultaneously through the MEC cluster of unit c. Summarizing all units in the system, we derive the network resource allocation vector Ys = {y s,1 , ..., y s,c The number of payloads of service provider s traversing the RAN domain under}:
[0107]
[0108] Finally, due to the number of concurrently executed jobs limited by the domain with the lowest performance, we can express the utility of the service provider as follows.
[0109]
[0110] 2. Network resource allocation method
[0111] We propose a techno-economic market model where service providers purchase resource packages in each edge node and cell individually, and then use these resources exclusively to execute jobs from their end users. In addition to the market equilibrium approach used for resource allocation, we also propose two approaches based on system performance optimization, which we will introduce one by one.
[0112] 2.1 Market equilibrium
[0113] A resource bundle is said to be and the corresponding price Will induce market equilibrium:
[0114] (i) Best Goods: Each service provider purchases only those goods that can produce the greatest unit of monetary utility, also known as “value for money”.
[0115] (ii) Market Clearing: The resource is fully allocated, or the corresponding price is zero. In addition, each service provider has exhausted its budget.
[0116] The first condition satisfies the rationality and selfishness of buyers, whose unique interest is to maximize the return on their market investment in the manner prescribed by their utility function. Intuitively, equilibrium cannot be established without this. Condition (ii) states that the offer for each resource can either meet the demand and be priced accordingly, or be in a surplus state and be given away for free. In addition, stability is only achieved when each buyer spends his entire budget, which means that any unused portion can indeed be spent to increase utility.
[0117] As long as the price does not exceed the budget, the service provider can purchase any resource bundle. Now it is clear that the balanced bundles are those that are most inclined towards the service provider.
[0118] Therefore, we develop a convex optimization procedure to find the market equilibrium for the above market model. Specifically, the formula is based on the above variables and appended with the following objective function (i.e., the first objective function):
[0119]
[0120] This first objective function maximizes the sum of the logarithms of the service provider utilities, weighted by the budget of each provider.
[0121] Our proposed mechanism works according to the following steps:
[0122] 1) Collect system model parameters and establish formulas;
[0123] 2) Optimize the model to obtain a balanced distribution and best SP utility
[0124] 3) Based on the above steps, allocate resources according to the entire system
[0125] Best Price and do not appear directly in the solution, but, due to their strong duality, they can be extracted as dual variables associated with the capacity constraint and . Therefore, they represent the rate of change of the objective function associated with any change in the capacity constraint. In other words, those resources whose capacity increase or decrease will have the largest impact on the objective have the highest prices, implying a direct proportional relationship between price and resource demand. Furthermore, prices will ensure that the cost presented by the optimal resource bundle for each provider is no higher than the provider's budget.
[0126] 2.2 Social Optimal Method
[0127] A linear optimization-based approach is proposed, which aims to allocate resources such that the total number of jobs that can be executed simultaneously without incurring performance degradation is maximized.
[0128] In this case, there is no guarantee that the resulting allocation represents any kind of market equilibrium. Therefore, the mechanism does not treat service providers as entities with decision-making capabilities, but rather as simply managing system resources according to some objective function to maximize performance.
[0129] In this problem, we propose two different objective functions to maximize the system performance: pure maximization of the sum of utilities (i.e., the total amount of work performed), and the sum of utilities weighted by the budget of each provider. These objectives, together with the previously defined variables, form the optimization model at the core of the proposed approach.
[0130] The second objective function, also called social optimality (SO), is defined as follows:
[0131]
[0132] This goal maximizes system performance regardless of budget differences between services. That is, it maximizes the number of served users.
[0133] In contrast, the third objective function is defined as:
[0134]
[0135] It includes a degree of fairness in the solution by giving more weight to service providers with larger spend, ensuring an advantage for such providers in the market model.
[0136] This solution is also called Weighted Social Optimum (WSO).
[0137] The resulting two allocation mechanisms (i.e., SO and WSO) follow the three-step approach of the market model. In other words, the optimization model is solved with the parameters of the underlying system and resources are allocated accordingly.
[0138] 2.3. Proportional sharing
[0139] The average value of allocating resources in a network slice and has been used as a baseline allocation scheme in resource management for network slicing and mobile edge computing. This scheme is based on a closed-form solution and does not employ any optimization techniques.
[0140] The proportional sharing (PS) allocation mechanism is that each service provider receives the capacity of each resource in proportion to the system budget. It can be regarded as an extension of the well-known generalized processor sharing algorithm in multi-resource scenarios.
[0141] make and The computing and network resource allocations of service provider s due to proportional sharing are respectively:
[0142]
[0143]
[0144] Once these quantities are calculated, the proportional sharing tool can be calculated using equation (3)
[0145] This allocation scheme takes neither the needs of individual providers nor the load profile of edge nodes and cells into account. In fact, it simply allocates all available resources so that service providers with larger budgets can get a relatively larger share. Since it does not reflect any changes in system conditions, it is considered a static allocation scheme. However, following the same principles of market-based models, this mechanism gives priority to those service providers with larger budgets.
[0146] Market model simulation scheme based on network system architecture
[0147] In simulations, we will numerically evaluate the performance and fairness of the aforementioned allocation methods by examining different instances applied to the system model.
[0148] Each market model is driven by the heterogeneity of resource demands among different buyers. It is less common for buyers to exhibit identical or similar demands, since the optimal solution is not far from an equal resource allocation. Therefore, in our numerical analysis, what matters are the ratios between resource demands among different service providers and relative to the cloud capacity, rather than their absolute values. This allows us to consider general computing equivalent units, CPU-U and memory equivalent units, respectively.
[0149] Therefore, the capacity It is used to measure the maximum available unit of each resource in a specific edge node. Similarly, the service requirements in the MEC domain Defines the units of computing resources required for a single job of the SP. Network resources are considered as a fraction of the total bandwidth available in a specific area. Therefore, the parameter and It is expressed in Hertz (Hz).
[0150] A simulation example was generated in MATLAB and the proportional sharing allocation mechanism was implemented in the same environment. The market equilibrium, social optimality and weighted social optimal mechanism were solved by IPOPT and CPLEX, respectively.
[0151] To represent the heterogeneity of resource requirements of different services, we define four service templates that represent possible demand configurations in cloud computing scenarios. In detail, we define three templates representing CPU-intensive, MEM-intensive, and bandwidth-intensive services, and another more balanced service with overall high resource requirements. These specific templates were chosen to facilitate the analysis of heterogeneous mixes of requirements. The values of the service templates are shown in Table 2:
[0152] Table 2
[0153]
[0154] Consider a localized MEC / RAN system consisting of heterogeneous network units and computing nodes. We set up nodes with different configurations in the RAN and MEC domains to simulate the system resource allocation process in various heterogeneous situations.
[0155] In each simulation run, one of the templates is randomly assigned to multiple service providers competing for system resources. In addition, Gaussian noise is added to each resource demand with a mean of 0 and a variance equal to 25% of the original value. This is done to account for demand oscillations caused by unpredictable changes in payload size, computational difficulty, and network conditions.
[0156] It should be noted that the method of the embodiment of the present disclosure can be performed by a single device, such as a computer or a server. The method of the present embodiment can also be applied in a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present disclosure, and the multiple devices will interact with each other to complete the described method.
[0157] It should be noted that the above describes some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0158] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present disclosure also provides a wireless network resource allocation device for an integrated energy service station.
[0159] refer to Figure 3 , a wireless network resource allocation device for an integrated energy service station, comprising:
[0160] The data transmission module 21 is used for at least one user end to transmit the data processing task to a network system model pre-constructed by connecting multiple edge nodes and multiple wireless units, wherein the edge nodes are network resources corresponding to data calculation and processing, and the wireless units are wirelessly connected to at least one server end and at least one user end respectively;
[0161] The data processing module 22 is used for at least one server to allocate appropriate edge nodes for data processing for each data processing task in the network system model according to the resource allocation strategy, wherein the predetermined resource classification strategy includes: market equilibrium strategy, social optimal strategy or proportional sharing strategy.
[0162] In a specific embodiment, the resource allocation strategy is a market equilibrium strategy;
[0163] The data processing module 22 specifically includes: a market equilibrium processing unit, specifically used for:
[0164] Determine the utility data u of each server s , and the resource budget value B required for the server to process each data processing task s ; Determine for each data processing task to satisfy the first objective function The corresponding edge nodes allocate data so that the corresponding data processing tasks can be processed using the edge node allocated data, where s represents each server and S is the total number of servers.
[0165] In a specific embodiment, the market equilibrium processing unit is further used for:
[0166] Determine the amount of work that each server executes simultaneously on the edge node side Determine the payload size of each server on the wireless unit side Determine the utility data for each server based on the minimum value of the maximum workload and the payload And, retrieve at least one resource budget score B used by each data processing task from the memory of the corresponding server. i , set at least one resource budget score B i Accumulate and obtain the resource budget value B corresponding to each data processing task s .
[0167] In a specific embodiment, the resource allocation strategy is a social optimal strategy; the data processing module 22 specifically includes: a social optimal processing unit, specifically used to:
[0168] Determine the utility data that can be executed simultaneously on each server s ; Using the second objective function Calculate the sum of utility data of at least one server, and determine the corresponding edge node allocation data for each data processing task based on the sum of utility data, so as to use the edge node allocation data to perform data processing on the corresponding data processing task, where s represents each server and S is the total number of servers.
[0169] In a specific embodiment, the social optimal processing unit is further used for:
[0170] Calculate the resource budget value B required by the server to process each data processing task s ; In determining the resource budget value B s When it is greater than a predetermined threshold, the third objective function is used The utility data sum is calculated, and the corresponding edge node allocation data is determined for each data processing task according to the utility data sum, so that the corresponding data processing task is processed by using the edge node allocation data.
[0171] In a specific embodiment, the resource allocation strategy is a proportional sharing strategy; the data processing module 22 specifically includes: a proportional sharing processing unit, specifically used to:
[0172] Determine the workload of the target server s' on the edge node side based on the resource ratio of the target server s' And, determine the payload size of the target server s' on the wireless unit side Determine the target server's utility data According to the utility data u of the target server s' s’ (X s’ , Y s’ ) distributes data to the target server s', so that the target server s' processes the data according to the data distribution result.
[0173] In a specific embodiment, the proportional sharing processing unit determines the workload of the target server s' on the edge node side according to the resource ratio of the target server s'. And, determine the payload size of the target server s' on the wireless unit side Specifically include:
[0174] Get the computing resource r capacity in edge node m and the network resource capacity in wireless unit c Determine the resource budget value B for each server s ; Calculate the resource budget value B of the target server s' s’ The proportion of: Determine the computing resource allocation for the target server s' and the market-based network resource allocation of the target server s' in, S is the total number of servers, M is the total number of edge nodes, R is the total amount of computing resources corresponding to an edge node, and C is the total number of wireless units. Determine the workload of the target server s' on the edge node side: Determine the payload size of the target server s' on the wireless unit side:
[0175] For the convenience of description, the above device is described by dividing it into various modules according to its functions. Of course, when implementing the present disclosure, the functions of each module can be implemented in the same or multiple software and / or hardware.
[0176] The device of the above embodiment is used to implement the corresponding integrated energy service station wireless network resource allocation method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.
[0177] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the integrated energy service station wireless network resource allocation method described in any of the above embodiments is implemented.
[0178] Figure 4 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.
[0179] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.
[0180] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.
[0181] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.
[0182] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).
[0183] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040).
[0184] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.
[0185] The electronic device of the above embodiment is used to implement the corresponding integrated energy service station wireless network resource allocation method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.
[0186] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the integrated energy service station wireless network resource allocation method as described in any of the above embodiments.
[0187] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.
[0188] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the wireless network resource allocation method for the integrated energy service station as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.
[0189] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Based on the concept of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present disclosure as described above, which are not provided in detail for the sake of simplicity.
[0190] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present disclosure difficult to understand, the known power / ground connections to the integrated circuit (IC) chips and other components may or may not be shown in the provided figures. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present disclosure difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present disclosure will be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present disclosure, it is apparent to those skilled in the art that the embodiments of the present disclosure can be implemented without these specific details or with changes in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.
[0191] Although the present disclosure has been described in conjunction with specific embodiments of the present disclosure, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.
[0192] The embodiments of the present disclosure are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the scope of protection of the present disclosure.
Claims
1. A method for allocating wireless network resources in an integrated energy service station, comprising: At least one user terminal transmits a data processing task to a network system model pre-constructed by connecting multiple edge nodes and multiple wireless units, wherein the edge nodes are network resources corresponding to data calculation and processing, and the wireless units are wirelessly connected to at least one server terminal and at least one user terminal respectively; The at least one server allocates appropriate edge nodes for data processing for each data processing task in the network system model according to a resource allocation strategy, wherein the resource allocation strategy includes: a market equilibrium strategy, a social optimal strategy, or a proportional sharing strategy; The resource allocation strategy is a market equilibrium strategy, and the at least one server allocates a suitable edge node for data processing for each data processing task in the network system model according to the resource allocation strategy, specifically including: Determine the utility data for each server u s , and the resource budget value required for the server to handle each data processing task B s ; Determine the first objective function for each data processing task The corresponding edge node allocates data, so as to use the edge node to allocate data to perform data processing on the corresponding data processing task, wherein s represents each server end, and S represents the total number of servers; The resource allocation strategy is a socially optimal strategy, and the at least one server allocates a suitable edge node for data processing for each data processing task in the network system model according to the resource allocation strategy, specifically including: Determine the utility data that can be executed simultaneously on each server u s ; Using the second objective function Calculate the utility data sum of the at least one server, and determine the corresponding edge node allocation data for each data processing task according to the utility data sum, so as to use the edge node allocation data to perform data processing on the corresponding data processing task, wherein s represents each server, and S represents the total number of servers; The resource allocation strategy is a proportional sharing strategy, and the at least one server allocates a suitable edge node for data processing for each data processing task in the network system model according to the resource allocation strategy, specifically including: According to the resource proportion of the target server s', determine the workload of the target server s' on the edge node side , and determine the payload size of the target server s' on the wireless unit side ; Determine the utility data of the target server s' ; According to the utility data of the target server s' obtained Data is allocated to the target server s', so that the target server s' processes the data according to the data allocation result.
2. The wireless network resource allocation method for an integrated energy service station according to claim 1, wherein: Determine the utility data of each server u s , and the resource budget value required for the server to handle each data processing task B s , including: Determine the amount of work that each server executes simultaneously on the edge node side ( X s ); Determine the payload size of each server on the wireless unit side ( Y s ); Determine the utility data for each server based on the maximum workload and the minimum value of the payload ;as well as, Retrieve at least one resource budget score used by each data processing task from the storage of the corresponding server B i , the at least one resource budget score B i Accumulate and obtain the resource budget value corresponding to each data processing task B s .
3. The wireless network resource allocation method for an integrated energy service station according to claim 1, wherein: Also includes: Calculate the resource budget required by the server to process each data processing task B s ; In determining the resource budget value B s When it is greater than a predetermined threshold, the third objective function is used The utility data sum is calculated, and corresponding edge node allocation data is determined for each data processing task according to the utility data sum, so as to perform data processing on the corresponding data processing task using the edge node allocation data.
4. The wireless network resource allocation method for an integrated energy service station according to claim 1, wherein: According to the resource proportion of the target server s', the workload of the target server s' on the edge node side is determined. , and determine the payload size of the target server s' on the wireless unit side , including: Get the computing resource r capacity in edge node m , and the network resource capacity in wireless unit c ; Determine the resource budget value B for each server s ; Calculate the resource budget value B of the target server s' s’ The proportion of: ; Determine the computing resource allocation of the target server s' , and the market-based network resource allocation of the target server s' , in, , S is the total number of servers, M is the total number of edge nodes, R is the total amount of computing resources corresponding to an edge node, and C is the total number of wireless units; Determine the workload of the target server s' on the edge node side: ; Determine the payload size of the target server s' on the wireless unit side: .
5. A wireless network resource allocation device for an integrated energy service station, comprising: A data transmission module, used for at least one user end to transmit a data processing task to a network system model pre-constructed by connecting multiple edge nodes and multiple wireless units, wherein the edge nodes are network resources corresponding to data calculation and processing, and the wireless units are wirelessly connected to at least one server end and at least one user end respectively; A data processing module, configured for the at least one server to allocate appropriate edge nodes for data processing for each data processing task in the network system model according to a resource allocation strategy, wherein the resource allocation strategy includes: a market equilibrium strategy, a social optimal strategy, or a proportional sharing strategy; If the resource allocation strategy is a market equilibrium strategy, the data processing module specifically includes: a market equilibrium processing unit, which is specifically used to: Determine the utility data for each server u s , and the resource budget value required for the server to handle each data processing task B s ; Determine for each data processing task to satisfy the first objective function The corresponding edge nodes allocate data so as to process the corresponding data processing tasks using the edge node allocated data, where s represents each server and S represents the total number of servers; The resource allocation strategy is a socially optimal strategy; the data processing module specifically includes: a socially optimal processing unit, specifically used to: Determine the utility data that can be executed simultaneously on each server u s ; Using the second objective function Calculate the sum of utility data of at least one server, and determine corresponding edge node allocation data for each data processing task according to the sum of utility data, so as to use the edge node allocation data to perform data processing on the corresponding data processing task, where s represents each server and S represents the total number of servers; The resource allocation strategy is a proportional sharing strategy; the data processing module specifically includes: a proportional sharing processing unit, which is specifically used to: Determine the workload of the target server s' on the edge node side based on the resource ratio of the target server s' , and determine the payload size of the target server s' on the wireless unit side ; Determine the utility data of the target server s' ; Based on the utility data of the target server s' Data is allocated to the target server s', so that the target server s' processes the data according to the data allocation result.
6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 4 when executing the program.
7. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 4.
